AI Visibility Analysis: Global Legal Industry

Ivica Srncevic | May 2026 | AI Visibility Research

The Firms That Argue for Everyone Else Cannot Argue for Themselves in AI

AI visibility in the legal industry measures how effectively a law firm’s digital presence is structured for retrieval, interpretation, and citation by AI systems, the engines now mediating the first moments of discovery for general counsels, procurement directors, CFOs, and the senior executives who choose outside counsel for the mandates that matter most.

These are not peripheral audiences. When a multinational’s GC is evaluating restructuring counsel, when a board’s audit committee shortlists firms for a significant regulatory matter, or when an in-house legal team is building a vendor panel for cross-border M&A, AI-assisted research is increasingly the first and most consequential stop. The firms that AI systems can parse with structural confidence appear in synthesized answers. The firms that they cannot parse structurally, regardless of deal flow, PEP score, or decades of precedent, do not.

This is the eighth instalment of my independent AI visibility research series. Previous analyses covered SaaS CRM, the world’s largest banks, industrial manufacturing, global life and health insurance carriers, the automobile industry, the commercial vehicle sector, and the global pharmaceutical industry. The legal industry now joins that dataset, and delivers a finding I did not expect: a sector defined by precision, persuasion, and structured argumentation that has, almost without exception, failed to structure its own digital presence for the systems now doing the searching.

Methodology

I evaluated each firm’s global corporate website using the AI Visibility Inspector and the Ivica Srncevic Frameworks. The assessment covers four structural dimensions that determine how confidently AI systems can retrieve, represent, and cite a brand:

• Structure – how content is architecturally organised for machine parsing
• Depth – the substantive quality and retrievability of content as AI systems process it
• Schema – the presence of structured data markup enabling confident entity identification
• Freshness – whether content age signals are present and verifiable to AI retrieval systems

The AI Retrieval Index runs from 0 to 100. Scores below 50 indicate significant structural invisibility. Scores between 50 and 74 represent fair to moderate visibility with material gaps. Scores at 75 and above indicate strong AI readiness.

The firms evaluated represent a cross-section of the world’s largest and most globally active law firms by revenue and headcount: DLA Piper, A&O Shearman, Gibson Dunn, Baker McKenzie, Kirkland & Ellis, Ropes & Gray, Sidley Austin, Skadden, White & Case, and Latham & Watkins.

The Scores

Sector average: 60.9 – Grade C, AI Retrieval Index

Zero firms in Grade A. Zero in Grade B. Nine in Grade C. One in Grade D. This is where global elite law stands in 2026 – performing marginally better than the pharmaceutical industry I analysed in the previous instalment, but still structurally failing an audience that increasingly begins its outside counsel search inside AI systems rather than rankings directories or peer referrals.

Five Findings the Legal Industry Cannot Afford to Ignore

Finding 1: DLA Piper Leads – But Leading a C-Grade Sector Is a Limited Distinction

DLA Piper scores 72, the highest in this dataset, and carries the sector’s strongest Schema score at 65 – a reading that stands meaningfully above every other firm. A&O Shearman follows at 71, with the strongest Freshness reading among the majority of the pack at 53 and a Schema of 50. These two firms have demonstrably invested more in structured signal implementation than their peers.

The critical context remains the same as in every sector I have analyzed: 72 still falls three points below the 75-point threshold that constitutes genuine AI readiness. DLA Piper leads an industry in which the best-performing firm has not yet crossed into Grade B territory. When GCs and procurement directors use AI to research outside counsel options, even the sector leader is not structurally optimised to win that retrieval moment.

The gap between sector leader and AI-ready is not a rounding error. It is a structural decision – one these firms have not yet made.

Finding 2: Structural Decay Is Universal – Every Single Firm

Every firm in this dataset triggered a Structural Decay warning. No exceptions. The causes, however, are instructive in revealing different failure modes.

Latham & Watkins and White & Case were both flagged for the same critical deficiency: no H1 tag found. An H1 is the primary semantic anchor for any web page – the signal that tells AI parsers what a page is about before any other content is processed. When it is absent, AI systems cannot anchor a primary topic. They are left to infer what the page represents from the surrounding, unstructured context. For firms whose homepage must communicate a clear, credentialed identity, global law firm, practice areas, and jurisdictional reach, that inference gap has direct consequences for how AI synthesises their representation in response queries.

Gibson Dunn presents a different failure mode: four H1 tags were found on a single page. Where Latham and White & Case offer AI parsers nothing to anchor to, Gibson Dunn offers too many competing signals at once. The result is fragmented intent, an AI system that cannot resolve which of the four claimed primary topics represents the firm’s actual identity. This is the digital equivalent of a firm responding to a brief with four different answers to the same question.

The remaining seven firms, Baker McKenzie, Kirkland & Ellis, Ropes & Gray, Sidley Austin, Skadden, A&O Shearman, and DLA Piper, triggered Structural Decay warnings for absent or unverifiable date signals. Their content age cannot be confirmed by AI retrieval systems. In practical terms, this means AI systems cannot determine whether a firm’s stated practice strengths, market positioning, or thought leadership content reflects current capability or an older iteration of the business.

This pattern has appeared in every sector I have analysed. In the legal industry, it carries a specific irony: these are organisations built on the precise dating of documents, the verifiable sequencing of filings, and the auditability of advice. The same rigour that governs a merger agreement has simply not been applied to the firm’s own public digital presence.

Finding 3: Schema Scores Reveal an Identity Problem at the Sector’s Core

The legal industry’s average Schema score is 36.5. DLA Piper leads at 65. A&O Shearman follows at 50. Every other firm in the dataset scores 35. This near-uniform floor is not coincidental; it reflects a sector-wide failure to implement the structured data markup that allows AI systems to identify what a firm is, what it does, where it operates, and what facts can be attributed to it with confidence.

Schema markup is the mechanism through which AI systems move from inference to identification. Without it, an AI system parsing Kirkland & Ellis’s homepage cannot definitively distinguish the firm as a specific legal entity from the broader category of private equity and restructuring counsel. It cannot attribute practice area leadership to the firm with structural confidence. It cannot confirm jurisdictional scope. It resolves all of this through unstructured text inference, a process that systematically advantages third-party sources, legal rankings platforms, and trade publications that carry better-structured signals than the firms themselves.

The practical consequence is visible in AI-generated answers to queries like “best law firms for leveraged buyouts” or “top firms for international arbitration.” The firms that appear are not necessarily those with the strongest track record. They are those whose digital presence is structured in a way that AI systems can parse and attribute with confidence. At a sector average Schema of 36.5, the firms in this dataset are consistently losing that retrieval competition to intermediaries.

DLA Piper’s Schema score of 65 and A&O Shearman’s score of 50 demonstrate that improvement is achievable and is not a function of firm size or content volume. It is a structural investment decision.

Finding 4: Freshness Is the Most Actionable Gap – and the Most Neglected

Four firms in this dataset scored in the single digits on Freshness: Baker McKenzie (4), Latham & Watkins (4), Ropes & Gray (8), and Sidley Austin (8). Skadden recorded 8. Kirkland & Ellis recorded 15. Six of ten firms are effectively invisible to AI systems in terms of content currency; their material cannot be date-verified, and AI retrieval systems treat unverifiable content as potentially stale regardless of its actual substance.

Gibson Dunn is the dataset’s most striking outlier on this dimension: a Freshness score of 89, the highest in this analysis and among the highest readings I have recorded across all sectors in this research series. This single score explains a significant portion of Gibson Dunn’s overall rating of 65, despite a Structure score of only 70 reflecting the H1 fragmentation issue. Gibson Dunn’s content is fresh, and AI systems can verify it. That structural confidence in currency translates directly to retrieval advantage for time-sensitive queries, new regulatory developments, recent deal announcements, and current practice area capability.

White & Case records a Freshness score of 57, the second-highest in the dataset and meaningfully above most peers. Again: this is a choice, not a technical constraint.

The implementation required to address Freshness failure across the sector is not complex. Adding dateModified JSON-LD to page templates is a development task measurable in days. The return is direct: AI retrieval systems gain the ability to confirm content currency, which shifts the confidence threshold for citation in the firm’s favour. The firms that have not done this are, in effect, allowing their content to compete at a structural disadvantage against legal news outlets, rankings publications, and client review platforms that routinely carry date signals their own sites do not.

Finding 5: Depth Is Consistently Strong – and Consistently Insufficient Alone

Nine of ten firms in this dataset score 75 or above on Depth. Latham & Watkins scores 85 despite recording the dataset’s lowest overall score of 47. A&O Shearman scores 85. Baker McKenzie, Kirkland & Ellis, and Ropes & Gray all score 85. The legal industry’s content depth, measured by the substantive quality and volume of retrievable material, is genuinely strong across the sector.

Latham & Watkins is the clearest illustration of the structural paradox this creates. A Depth score of 85, excellent by the standards of this research series, paired with a Structure score of 55 and a Freshness score of 4, produces an overall AI Retrieval Index of 47: Grade D, the only firm in this dataset below the Grade C threshold. Latham & Watkins publishes substantive content. AI systems cannot reliably anchor that content to the firm’s identity, cannot verify its currency, and cannot parse the homepage to understand what the firm’s primary topic is. The content exists. The structural signals that would allow it to be cited with confidence do not.

This pattern is the most consistent finding across all eight sectors in this research series: content depth cannot compensate for structural signal failure. It is a necessary condition for AI visibility. It is not sufficient.

The Legal-Specific Risk: When AI Cannot Distinguish Counsel from Category

Every industry faces commercial consequences from AI invisibility. The legal industry faces a dimension that amplifies those consequences in a specific way: it competes for mandates that are relationship-driven, referral-dependent, and resistant to marketing in traditional forms, which makes AI retrieval disproportionately influential in the moments where it does operate.

When a senior in-house lawyer uses an AI system to research outside counsel for a cross-border regulatory matter, the AI response is not merely a list of options. It is a synthesised representation of capability, credibility, and relevance, assembled from whatever structural signals the AI system can parse and attribute with confidence. The firms that win that moment are not necessarily the ones with the strongest track record on the matter type. They are the ones whose digital presence provides AI systems with the structural confidence to cite them specifically, rather than citing the practice area category generally.

A firm that allows AI systems to represent it as “a global law firm”, the generic entity classification that unstructured parsing produces, has surrendered the specific competitive positioning that its practice group structure, jurisdictional depth, and sector specialisation should provide. It is not a neutral outcome. It is a structural loss to better-optimised intermediaries, directories, and competitors.

The firms that address these structural gaps earliest are not simply improving their search performance metrics. They are asserting control over how AI systems represent their identity to the audiences that now begin, and increasingly complete, their outside counsel research process inside AI interfaces rather than rankings tables.

What AI Actually Sees

The AI assessment outputs from the AI Visibility Inspector reveal, directly, what AI systems extract when they parse these pages.

DLA Piper’s site was identified as being about a “Global Law Firm”, the broadest possible entity classification, with no practice differentiation, no sector specificity, and no jurisdictional granularity. A firm of DLA Piper’s global reach, present in 40-plus countries, with depth across real estate, corporate, litigation, and regulatory, is being represented by AI systems at the level of a category label. The Schema score of 65, the highest in the dataset, has not yet bridged the gap from category to specific entity.

A&O Shearman was parsed as “A&O Shearman, global law firm.” The inclusion of the firm name in the entity description is a material improvement over pure category attribution; it indicates that AI systems can anchor the firm as a distinct entity rather than a generic representative of the practice type. The Schema score of 50 is producing a measurable identity benefit relative to the sector norm.

The firms scoring 35 on Schema, Baker McKenzie, Kirkland & Ellis, Ropes & Gray, Sidley Austin, Skadden, Gibson Dunn, White & Case, and Latham & Watkins, are all operating below the threshold at which AI systems can differentiate them with structural confidence. The commercial significance of that differentiation gap grows with every AI system that replaces or precedes a directory search or peer referral.

The Sector Average in Context

The legal sector performs in line with banking, both Grade C sectors with similar structural failure patterns. The common denominator across every sector in this research series is the same: Schema implementation is the universal gap, Freshness is the most neglected dimension relative to implementation cost, and Depth is consistently the strongest score despite being insufficient alone.

What distinguishes the legal sector is the specific nature of the structural irony. These are organisations whose core product is structured argumentation, the precise construction of documents, filings, and communications designed to be unambiguously parsed by their audience. They apply that discipline to client work as a matter of professional obligation. They have not applied it to their own digital presence as a matter of competitive strategy.

Recommendations

For all ten firms:

The structural improvements that would move these scores into Grade B territory are not strategically complex. They require execution, not insight.

Implementing dateModified JSON-LD across page templates is a development task that directly addresses the Freshness gap afflicting eight of ten firms. The commercial return, AI retrieval systems gaining the ability to confirm content currency, is immediate and measurable.

Auditing and consolidating H1 tag architecture addresses the structural decay warnings affecting Latham & Watkins, White & Case, and Gibson Dunn. A single, clear primary H1 per page is the baseline condition for AI parsers to anchor topic identity. These are not redesign projects. They are structural corrections.

Expanding Schema markup to include LegalService entity types, jurisdictional coverage, and practice area classification is the work that transforms entity inference into entity identification. DLA Piper’s Schema of 65 demonstrates the ceiling that is available. The sector average of 36.5 demonstrates how far most firms are from reaching it.

For firms below 60:

Latham & Watkins, Skadden, Sidley Austin, and White & Case should treat the AI Retrieval Index as a diagnostic with commercial implications, not a technical metric. A Depth score of 85 publishing into an AI ecosystem that cannot anchor, date, or specifically identify its source is a structural waste of content investment. The corrections are well within the technical capability of any firm’s digital team.

Conclusion: The Firms That Cannot Be Found Cannot Be Hired

The legal industry has spent decades building brands through relationships, rankings, and reputation. These channels remain important. They are no longer sufficient as the only channels that matter.

AI systems are now the first interface for a growing proportion of research conducted by the audiences these firms most need to reach. The firms that appear in AI-synthesised answers to outside counsel queries, confidently identified, specifically attributed, and structurally credible, gain a discovery advantage that compounds over time as AI usage in professional research increases. The firms that do not appear to surrender that moment to better-structured intermediaries, to competitors who have made different structural decisions, and to the category-level representations that AI systems produce when they cannot parse a specific entity with confidence.

The sector average of 60.9 is not a catastrophe. It is a baseline, and baselines have competitors. The first firm in this dataset to cross 75 on the AI Retrieval Index will not simply improve its own score. It will establish what elite legal AI visibility looks like and create the reference point against which every other firm in the sector will eventually be measured.

That firm has not yet emerged from this dataset. The opportunity remains open.

This research is part of an ongoing independent series analyzing AI visibility across global industries. Previous installments cover the SaaS CRM sector, global banking, industrial manufacturing, life and health insurance, the automobile industry, the commercial vehicle sector, and the global pharmaceutical industry. All assessments use the AI Visibility Inspector and the Ivica Srncevic Framework.

Ivica Srncevic is an independent SEO and AI visibility strategist. This research is conducted independently and is not sponsored or commissioned by any of the firms assessed.